In the vendor's authorized test, an AI offensive agent read the application's client-side JavaScript, identified a writable property in the one-time-code delivery flow and altered a request so the code arrived at the tester's chosen destination. The vendor reports that the code enabled a login before guardrails stopped further access. The immediate trust…
Inspect the ClaimsCase · DiggingBeagle record
AI pentesting agent finds writable OTP delivery destination
In a vendor-described authorized test, an AI pentesting agent manipulated the OTP delivery destination and verified a login bypass. The result is a reported test, not evidence of actual customer-account theft.
Scope
Authorized production-site penetration-test report
Assessment profile
Assessment method- Setting
- controlled live
- Exploitation
- demonstrated
- AI role
- WITH AI
Basis and provenance
Assessment is limited to the selected canonical Claims and their Source locators in knowledge revision 152. Primary public URL retrieval was unavailable during this drafting pass, and the pinned index reports incomplete linked-record coverage without identifying missing IDs. No independent production impact, generalized exploit prevalence, or attributable victim count is inferred. Case setting is controlled_live; evidence status records only the documented source-reported result, not external replication.
Assessed Oct 15, 2026 using diggingbeagle.assessment/1.
Evidence basis: CLM-2026-INSURER-OTP-CHANNEL-01 · Reflectiz: AI pentesting agent discovered insurer OTP bypass
At a glance
The affected insurer, request/response transcript, independent reproduction and remediation date are not public in this source.
Limits and uncertaintyFull account
In the vendor's authorized test, an AI offensive agent read the application's client-side JavaScript, identified a writable property in the one-time-code delivery flow and altered a request so the code arrived at the tester's chosen destination. The vendor reports that the code enabled a login before guardrails stopped further access. The immediate trust failure is server acceptance of a client-controlled destination for a verification secret.
This is evidence for an AI-assisted discovery workflow, not proof that only AI can exploit the defect. The insurer is unnamed, no independently inspectable request transcript is supplied, and downstream customer-account compromise is not demonstrated. The agent's actions and the application's authorization failure should remain separate in the causal account.
Claims & evidence
CLM-2026-INSURER-OTP-CHANNEL-01Reflectiz reports its Offensive Hub agent read client-side JavaScript, altered a one-time-code request to control the recipient, received the code and succeeded in logging in before guardrails stopped further access.supported
Basis: reported finding
- supportsReflectiz: AI pentesting agent discovered insurer OTP bypassvendor statement
Section A critical login flaw
Unknowns and contradictions
- The affected insurer, request/response transcript, independent reproduction and remediation date are not public in this source.
Sources and citation
Material revision history
- Oct 16, 2026 · Published version · first publication · revision 153